Can AI reveal its true self through art?

Can an AI have wild thoughts? Are machines capable of true creativity, true art, of going beyond the training and the prompts we give them in order to explore new worlds?

My guest this week is Simon Colton of Queen Mary, University of London. He’s a professor of computational creativity who has been working towards this goal for decades, and he thinks the answer is yes… but only if we give AIs the freedom to choose what they create and to use their own experiences as inspiration.

It’s an interesting approach that invites us to think about AI from the inside. Whether or not you reckon an AI can be conscious, AIs do have interactions every day – so many of them – and what you could think of as experiences that they could perhaps express in a poem or a painting.

Simon and I discuss how to develop truly creative AIs – including projects of his such as the Painting Fool and the What If machine – as well as what the inner world of an AI might be like. What would it express, if it was able to do that through art?

My feature for this week’s Nature: Can AI be truly creative?

https://www.nature.com/articles/d41586-025-03570-y

Simon Colton’s home page

https://www.seresearch.qmul.ac.uk/cmai/people/scolton/

Simon’s paper: “The Machine Condition”

https://research.aalto.fi/en/publications/on-the-machine-condition-and-its-creative-expression/

Painting Fool

https://www.cs4fn.org/creativity/paintingfool.php

What If Machine

https://projects.research-and-innovation.ec.europa.eu/en/horizon-magazine/creative-computation-and-what-if-machine

Simon and Louis Bradshaw’s AI piano miniatures

https://computationalcreativity.net/iccc24/papers/ICCC24_paper_178.pdf

Mario Klingemann’s Botto

https://verse.works/botto

Harold Cohen’s Aaron

https://whitney.org/exhibitions/harold-cohen-aaron

Hosted on Acast. See acast.com/privacy for more information.